Re: [PATCH] mm,thp,shmem: limit shmem THP alloc gfp_mask

[Michal Hocko]

On Wed 21-10-20 23:48:46, Rik van Riel wrote:
> The allocation flags of anonymous transparent huge pages can be controlled
> through the files in /sys/kernel/mm/transparent_hugepage/defrag, which can
> help the system from getting bogged down in the page reclaim and compaction
> code when many THPs are getting allocated simultaneously.
> 
> However, the gfp_mask for shmem THP allocations were not limited by those
> configuration settings, and some workloads ended up with all CPUs stuck
> on the LRU lock in the page reclaim code, trying to allocate dozens of
> THPs simultaneously.
> 
> This patch applies the same configurated limitation of THPs to shmem
> hugepage allocations, to prevent that from happening.
> 
> This way a THP defrag setting of "never" or "defer+madvise" will result
> in quick allocation failures without direct reclaim when no 2MB free
> pages are available.

I remmeber I wanted to unify this in the past as well. The patch got
reverted in the end. It was a long story and I do not have time to read
through lengthy discussions again. I do remember though that there were
some objections pointing out that shmem has its own behavior which is
controlled by the mount option - at least for the explicitly mounted
shmems. I might misremember.

[...]

> diff --git a/mm/shmem.c b/mm/shmem.c
> index 537c137698f8..d1290eb508e5 100644
> --- a/mm/shmem.c
> +++ b/mm/shmem.c
> @@ -1545,8 +1545,11 @@ static struct page *shmem_alloc_hugepage(gfp_t gfp,
>  		return NULL;
>  
>  	shmem_pseudo_vma_init(&pvma, info, hindex);
> -	page = alloc_pages_vma(gfp | __GFP_COMP | __GFP_NORETRY | __GFP_NOWARN,
> -			HPAGE_PMD_ORDER, &pvma, 0, numa_node_id(), true);
> +	/* Limit the gfp mask according to THP configuration. */
> +	gfp |= __GFP_COMP | __GFP_NORETRY | __GFP_NOWARN;

What is the reason for these when alloc_hugepage_direct_gfpmask provides
the full mask?

> +	gfp &= alloc_hugepage_direct_gfpmask(&pvma);
> +	page = alloc_pages_vma(gfp, HPAGE_PMD_ORDER, &pvma, 0, numa_node_id(),
> +			       true);
>  	shmem_pseudo_vma_destroy(&pvma);
>  	if (page)
>  		prep_transhuge_page(page);
> 
> -- 
> All rights reversed.

-- 
Michal Hocko
SUSE Labs